I do wish I had documented what I tried better! There might be a magic combo that could have helped but I tried tweaking a lot of the vm settings.
One day I will probably see if I can still reproduce the original problem and be more methodical about it. More likely on list of things I might not ever get around to.
3. vm.oom_kill_allocating_task is a footgun. It kills the last task that asked for memory and it could be any random task in the system.
4. disabling overcommit is another footgun, it makes malloc fail long before the memory is exhausted. See for a detailed explanation https://unix.stackexchange.com/a/797888/1027
reply